01 / The complete workflow
From search data to a reviewed website change
Choose a page, collect the evidence, brief your assistant and check what it produces.
Search and website practice
Make your next website change worth making.
Use search evidence to decide what to improve, then give your AI assistant a brief you can inspect. Start with one page and carry its sources through to the finished change.
Read the workflowGuides and blank working templates. No paid account is needed to prepare a brief.
Choose your task
Read, then make something useful.
02 / Data access
Connect Semrush MCP with a bounded research task
Check access, discover reports and preserve the result. Documentation reviewed; paid-account execution is untested.
03 / Measurement
Investigate AI citations with a repeatable question panel
Keep mentions, citations, recommendations and referrals distinguishable.
04 / Content relationships
Connect pages and the evidence behind them
Use a content graph when it improves retrieval and maintenance decisions.
Choose data for the decision.
Your search performance
Use the first-party reports available for your website. Record the period and missing coverage before interpreting a trend.
Market estimates
Evaluate Semrush or DataForSEO for a bounded research task. Keep provider estimates distinct from your own site's results.
Page content
Use Firecrawl, a specified Apify Actor or a self-hosted crawler to inspect pages. Extraction alone cannot tell you search demand.
These are editorial options. Links in this hub do not establish a partner relationship, account access or a tested comparison. See the affiliate policy for monetized recommendations elsewhere.
Free working files
Keep the method. Use your own data.
Download the blank files directly. They contain no measured results, account keys or hidden installation steps.
- Content brief
Reader, evidence, permissions and acceptance in one editable file.
content-brief.md
- Evidence record
A blank record that keeps missing values explicitly unknown.
evidence-record.json
- Measurement log
A spreadsheet-ready header for your own observations.
measurement-log.csv
- Experiment protocol
Define the comparison, observation window and stop rule.
experiment-protocol.md
Your website
Give your assistant a better starting point.
Prepare a brief from the evidence you already have. This form runs in your browser and does not send field values to a server. Clear the form when you finish.
Experiments worth repeating.
Use these protocols to test your own workflow. They are proposed comparisons; this page reports no winning tool or search uplift.
Prompt-only research versus a sourced brief
Give both arms the same task, model and time allowance. Change only the evidence access. Review source support, accepted output, cost and human correction time. Record failures as well as successes.
Hosted extraction versus a self-hosted crawler
Freeze a permitted URL set and manually label the required fields. Compare accuracy, missing fields, retries, runtime cost and maintenance effort. Hold the demand-data source constant.